STNetEncoder

class selfeeg.models.encoders.STNetEncoder(Samples, F: int = 256, kernlength: int = 5, dropRate: float = 0.5, bias: bool = True, seed: int = None)[source]

Pytorch implementation of the STNet Encoder.

See STNet for some references. The expected input is a 4D tensor with size (Batch x Samples x Grid_width x Grid_width), i.e. the classical 2d matrix with rows as channels and columns as samples is rearranged in a 3d tensor where the first is the Sample dimension and the last 2 dimensions are the channel dim rearrange in a 2d grid. Check the original paper for a better understanding of the input.

Parameters:

  • Sample (int) – The number of EEG Samples.

  • F (int, optional) –

    The number of output filters in the convolutional layer.

    Default = 8

  • kernLength (int, optional) –

    The length of the convolutional layer.

    Default = 5

  • dropRate (float, optional) –

    The dropout percentage in range [0,1].

    Default = 0.5

  • bias (bool, optional) –

    If True, adds a learnable bias to the convolutional layers.

    Default = True

  • seed (int, optional) –

    A custom seed for model initialization. It must be a nonnegative number. If None is passed, no custom seed will be set

    Default = None

Example

>>> import selfeeg.models
>>> import torch
>>> x = torch.randn(4,128,9,9)
>>> mdl = models.STNetEncoder(128)
>>> out = mdl(x)
>>> print(out.shape) # shoud return torch.Size([4, 1296])
>>> print(torch.isnan(out).sum()) # shoud return 0